Lies, Damn Lies, And Statistics Data Visualisation-ppbox

Presentation 1. Chose the right graph type for your data trends are best presented as line graphs, whilst totals are best as histograms (bar charts). Proportions are best .pared using pie charts, and lots of sets of similar data, lends itself to tables. 2. Titles give your tables, graphs and charts easy to understand titles, and make sure they really explain what the data being presented is all about. Instead of Graph of population say Graph showing exponential population increase over the last 5 years. 3. When deciding whether to present your data in a table or a graph, consider that graphs are good at quickly showing rough trends, they dont necessarily require detailed or specific data, and are good at showing significant .parisons. Tables, however, are very good for showing large amounts of data, exact figures, and lots of sets of numbers. 4. Our eyes find it easy to scan the horizontal and vertical rows of tables and you can let the data itself guide the eyes. You dont need horizontal or vertical lines (rules), which can even get in the way of scanning. Consider using tinted lines behind rows or columns of figures to distinguish them, and when column headings are long words or sentences, consider running them on to two or more lines. You can also run them diagonally, or if absolutely necessary, abbreviate the words to make them shorter. Abbreviation works to really good effect in football league tables where the data content is so well known that the column headings W L P F for example are immediately read as Won, Lost, Played, Goals for etc. 5. Keep design simple and dont use fancy or elaborate images of things to represent data. If you use pictures of things (like icons) to present statistics, dont make one pictogram bigger than another because the eye cant detect how much bigger one is from another. Instead, use one pictogram to stand for one unit of value, and have multiple pictograms to represent .parable sets of data. (This is often known as Isotype, and is named after the Istotype institute who pioneered this method of graphic statistics.) 6. Make sure that scales are the same i.e. consistent across a series of charts if you dont, a reader will .pare two graphs thinking that they are visually .parable when they are not. Changing the scale just confuses and could even deceive readers. About the Author: 相关的主题文章: